Write a Python function that reads two CSV files – patients.csv (patient_id, name, dob) and labs.csv (patient_id, visit_date, test_name, result_value, unit, reference_range_low, reference_range_high) – and returns a DataFrame of patients who had a Hemoglobin result below the lower limit of normal (LLN) in any visit, including visit_date and result_value. Handle missing values and duplicate rows.

Use the CIRCLES framework: Clarify, Identify, Recommend, Create, List, Execute, Summarize. 1) Clarify inputs/outputs. 2) Identify key steps: load CSVs, drop duplicates, handle NaNs, filter Hemoglobin, compare to LLN, select columns. 3) Recommend vectorized Pandas ops. 4) Create function skeleton. 5) List edge cases. 6) Execute code. 7) Summarize return type. 120‑150 words.

First, I import pandas and define the function signature: def get_low_hemoglobin(patients_path, labs_path). I read both CSVs using pd.read_csv, specifying dtype for patient_id to avoid type mismatches. I drop duplicate rows in labs using drop_duplicates on ['patient_id', 'visit_date', 'test_name']. I handle missing values by dropping rows where result_value or reference_range_low is NaN, as they cannot be evaluated. I filter labs for test_name == 'Hemoglobin', then create a boolean mask where result_value < reference_range_low. I merge the filtered labs with patients on patient_id to bring in patient demographics. Finally, I select patient_id, name, visit_date, result_value, and return the resulting DataFrame. I include a docstring explaining parameters and return type. This solution is fully vectorized, handles edge cases, and is ready for unit testing.